Genetic diversity and natural selection of Plasmodium vivax multi-drug resistant gene (pvmdr1) in Mesoamerica

All infected blood samples were obtained from symptomatic patients seeking malaria diagnosis in Mexico and Nicaragua (in each case, after informed consent was given). From 2008 to 2010 in Mexico, 93 samples were obtained in the laboratory facility at the Regional Research Center for Public Health (CRISP-INSP) from patients living in the Tapachula municipality and its surroundings in southern Chiapas (SMX), the southernmost point of Mexico that borders with Guatemala [16]. From 2011 to 2012 in Nicaragua, 107 samples were selected from the sentinel laboratory network established by the Health Ministry at head municipalities, 83 from the Northeast (NIC-NE; RACCN, North Caribbean Coast Autonomous Region) and 24 from the Northwest (NIC-NW; North Pacific Coast). Another 14 samples were included that had been obtained in NIC-NW during 2006–2007 [17]. Distances between these sites are the following: ≈611 km from SMX to NIC-NW, ≈853 km from SMX to NIC-NE, and ≈331 km from NIC-NE to NIC-NW. The NIC-NE in comprised by the Miskito and the mining regions.

The diagnosis of P. vivax was carried out by microscopic examination of stained thick blood smears, coming from capillary blood samples (before treatment) used to impregnate filter paper (Whatman #2). Then the participants were administered the CQ-PQ combination treatment, which is in accordance with health standards in Mexico [18, 19] and in Nicaragua [20]. Genomic DNA was extracted from P. vivax-infected bloods impregnated on filter paper, using a commercially available QIAmp DNA blood Minikit (Qiagen, USA), and following the manufacturer’s instructions.

A gene fragment of about 600 bp and containing codon positions 976 and 1076 was amplified using the following oligonucleotides: F3.2 5′˗ACC AGG ATA GTC ATG CCC˗3′ (nt 2747–2764) and R3 5′˗TCT CCC TTT AGG GAC ATC AAC˗3′ (nt 3384–3368). The PCR reaction was prepared as follows: 10 μL of 5× PCR buffer, 5 μL of MgCl₂ (25 mM), 2.5 μL of the dNTPs (1.25 mM), 2.5 μL of each primer (10 μM), 0.5 μL of Go Taq DNA polymerase, 5 u/μL and 2–4 μL of sample DNA for a final volume of 50 μL. The PCR reaction conditions were as follows: 5 min at 94 °C followed by 35 cycles: 1 min at 94 °C, 1 min at 60 °C, and 1 min at 72 °C; afterwards, there was a final extension of 72 °C for 10 min.

To investigate the presence of other nucleotide changes reported previously [6], two additional nucleotide fragments were PCR amplified. To that, primers F1 5′˗GAG GGA GAT GTC ATC ATC AAC GA˗3′ (nt 1315–1337) and R1 5′˗CTT CTG TCC ACC TGA CAA CTT AG˗3′(nt 1733–1755), and F2 5′˗CAA GGA CAG CAA TGA GAA GAA˗3′ (nt 2013–2033) and R2 5′˗CGC GAT GAC TAA GAT GTA GAG G˗3′ (nt 2601–2622) were used. The PCR reaction was prepared as follows: 10 μL of 5× PCR buffer, 4 μL of MgCl₂ (25 mM), 2.5 μL of dNTPs (1.25 mM), 1.9 μL of each primer (10 μM), 0.5 μL of Go Taq DNA polymerase 5 u/μL (Invitrogen Corporation, Carlsbad, CA), and 2–4 μL of extracted DNA for a final volume of 50 μL. The PCR reaction conditions were as follows: 3 min at 94 °C followed by 35 cycles: 40 s at 94 °C, 40 s at 54 °C, and 1 min at 72 °C; afterwards, there was a final extension of 72 °C for 5 min. All PCR reactions were run in a MyCycler (BioRad, Hercules, CA, USA).

All amplified products were resolved in agarose gels at 1%, and stained with 0.2 µg/mL ethidium bromide using an electrophoresis chamber Midicell primo (Thermo EC330, New York, USA). A molecular marker of 100 bp ladder was used (Invitrogen Corporation, Carlsbad, CA, USA). PCR products were then purified using a MiniElute PCR Purification Kit (Qiagen, Valencia, CA, USA), according to the manufacturer’s instructions. The purified products were Sanger sequenced by using forward and reverse primers (at the High Throughput Genomics Unit, Department of Genome Sciences, University of Washington, Seattle, WA, USA).

The quality of pherograms with the forward and reverse nucleotide sequences was verified manually and by using BioEdit v7.1.3 software. DNA sequences with rare polymorphisms were re-amplified and sequenced. All pherograms showed a single genotype pattern. The consensus sequences obtained for each gene fragment were submitted to the NCBI-Gen Bank [Accession Numbers: KX180164–KX180638].

The Salvador I strain (Sal I) mdr1sequence was used as reference: XM_001613678. Sequences forward and reverse were aligned using ClustalW Multiple Alignment of BioEdit v7.0 [21] and revised manually. Nonsynonymous and synonymous changes were identified, and the frequency of each nucleotide change was calculated per geographic site in Mesoamerica (NIC-NE, NIC-NW and SMX), as well the haplotypes were constructed to background mutations at codons 976/1076 with the dnaSP program v5.1 [22]. Statistical analysis was carried out with STATA v12.1.

In order to explore the regional and global parasite relationships based on pvmdr1, haplotype networks were constructed using TCS 1.21 [23]. Mutational steps represent the connections between haplotypes, and empty squares showed the non-sampled or extinct haplotypes. The colour of the circles represents the geographic origins of each haplotype, while the size of the circle represents the frequency of each haplotype.

Indexes of nucleotide diversity were calculated in dnaSP v5.1. To test whether natural selection and gene flow shaped the evolution of pvmdr1 gene in parasite populations of Mesoamerica, the number of synonymous (s) and nonsynonymous (ns) nucleotide changes and the difference in the rate of nonsynonymous versus synonymous changes (dN˗dS) were determined by using the Nei Gojobori proportion method with 1000 bootstrap replicates in MEGA v6.0 software [24]. Tajima’s D test and the minimal number of recombination events were calculated using dnaSP. To estimate the P. vivax genetic relationships from different locations, the F _ST values were estimated by using the two parameters of Kimura in dnaSP; values ranged from 0.0 to 1.0. A value of 0.0 indicates that populations are equal and 1.0 indicates they are completely different [25, 26].

DNA sequences for the global genetic analysis. Groups of DNA sequences that comprised codons 925–1083 (including positions 976/1076) were obtained from the NBCI Gene Bank. Brazil (BRZ): n = 7, EU333973-9 [5]; n = 3, AY571981-3 and 79–83, 85–89 [14]; n = 15, KM016495/502, 04, 05 and 07–17 [10]. Papua New Guinea (PNG): n = 6, AY571975-80 [14]. Iran (IR): n = 4, KM216181-4 (Sharifi-Sarasiabi et al. unpublished). Madagascar (MD): n = 80, EU683813-19 [14]. South Korea (SK): GU476519 (Chen et al. unpublished). India (IND): n = 49, KC818349-78, 80, 82, 85, 87, 89, 90, 92, 93, 95, 96, 98, 99/400, 02, 04, 06–09, 11 and 12 [27]. Cameroon: n = 8, KJ534638-45 (Ngasa et al., pers.comm.). Additionally, other DNA sequences were reconstructed, using the aforementioned sequence reference of the Sal I strain as a template (AY618622, with codon 958 atg/M; AY571984 and XM001613678 with codon 958 acg/T), and only if the length and coordinates of the pvmdr1 gene fragment, synonymous and nonsynonymous nucleotide changes, and haplotype frequency were clearly described: IND, n = 25 [28]. Honduras (HON), n = 37 [13]. Nepal (NEP), n = 39 [29]. Ecuador (ECU), n = 17; Sri Lanka (SLK), n = 119; Pakistan (PK), n = 24; Sudan (SUD), n = 4; Sao Tomé (SAT), n = 3 [11].

Upon analysing 1536 bp of pvmdr1, two synonymous and six nonsynonymous nucleotide changes were detected (Table 1). From 163 parasites (66, 23 and 74 from NIC-NE, NIC-NW and SMX, respectively), two changes (one nonsynonymous and one synonymous) were detected in the gene fragment of 411 bp (nucleotides 1330–1740). The nonsynonymous change was observed (only in NIC-NW) at codon 500 D → N). While the synonymous change at codon 529, was highly frequent in SMX and NIC-NE, but not detected in parasites from NIC-NW. Interestingly, from 155 parasites (65, 26 and 64 from NIC-NE, NIC-NW and SMX, respectively) a second fragment of 549 bp (nucleotides 2767–3315) was obtained, and six nucleotide changes were detected. The nonsynonymous changes at codons Y976F and F1076L were highly frequent in NIC-NE and NIC-NW, but not detected in SMX. Similarly, the synonymous change at codon 1021 was highly frequent in NIC-NE and NIC-NW, but rare in SMX. The change at codon A927T exclusive to NIC-NE, it was detected in parasites from different municipalities of the Miskito and the mining regions. The substitution T958M was highly frequent at the three sites, while the F1070L change was detected in only two isolates from NIC-NE (Rosita and Prinzapolka municipalities) and two isolates from SMX (from the year 2010). The third gene fragment of 576 bp (nucleotides 2032–2607), obtained from 148 parasites (63, 17 and 72 from NIC-NE, NIC-NW and SMX, respectively), was identical to the Sal I strain sequence.

There were 141 parasite isolates (59 from NIC-NE, 59 from SMX and 23 from NIC-NW) with the nucleotide information at variant codons 500, 529–927, 958, 976, 1021, 1070 and 1076 resolved 13 pvmdr1 haplotypes (m1 → m13; ordered by their frequency) (Fig. 1). There were differences in haplotype frequencies between the three sites (P = 0.000): haplotype m1 GG-GTACTT was predominant (93.2%) in SMX; m3 AA-GTTTTC was highly frequent (78.2%) in NIC-NW; m2 GA-GTTTTC and m4 GG-ATATTC were at 62.7 and 22%, respectively, in NIC-NE. Other haplotypes were detected in two or all sites, although at a very low frequency. For instance, haplotype m1 was also detected in one isolate from NIC-NE and in another from NIC-NW, m6 and m10 were observed at SMX and NIC-NE, and m5 was found in NIC-NW and NIC-NE. One haplotype (m8) was exclusive to SMX and resembled the Sal I strain sequence. Another three haplotypes (m9, m11 and m13) were exclusive to NIC-NE and two (m7 and m12) to NIC-NW (Fig. 1). There were four haplotypes having nucleotide changes at codons 976F (a → t) and 1076L (t → c); m2, m3, m7: GG-GTTCTC, m12: GG-GTTTTC; those haplotypes differed among them by one to three mutations. Two other haplotypes showed only 1076F/L; m4 and m11: GA-ATATTC. The haplotypes m2 and m4, highly frequent at NIC-NE were also frequent in all sampled municipalities: Rosita, Bonanza, Waspam and Puerto Cabezas. The higher number of exclusive haplotypes were detected in Waspam (n = 3), the Miskito region that borders Honduras.

The haplotype network of P. vivax mdr1 comprising mutations at codons 500, 529–927, 958, 976, 1021, 1070 and 1076 for parasites from the study sites is shown in Fig. 2. All 13 haplotypes were found closely related; most haplotypes were at one mutational step from each other, and at 1–9 mutational steps among them. Haplotypes from SMX were only separated by 1–2 mutational among them. The most frequent haplotype (m1) in SMX was at one mutational step from the Sal I strain sequence. While haplotypes m2 and m3 were at about 6–9 mutational steps from m1; they were highly frequent in NIC-NE and NIC-NW, respectively.

Parasites from NIC-NE and mostly from the Mesoamerican region showed a nucleotide diversity (π = 0.0036) similar to that found at other geographic origins as well as the global level (π = 0.0028). Moreover, the minimal number of recombination events were from very low to cero (the value for Mesoamerica was Rm = 2) (Table 2).

Tajima’s D test statistical value was negative for SMX and NIC-NW, while being positive for NIC-NE and other sites, but no significant. The lowest value was for Brazil (−1.681) and the highest for Ecuador (0.949), but in either case, it was found to be no significant. Although a higher number of nonsynonymous than synonymous changes were detected, the dN–dS values showed no significant departure from neutrality. Parasites from Nepal had the highest dN–dS value with marginal significance (1.845; P = 0.067) (Table 2).

When using global sequences of P. vivax mdr1, the haplotype network evidenced the relationship of 19 global haplotypes (g1 → g19) identified in 610 sequences (Fig. 3). The most frequent haplotype g1 and probably the ancestral haplotype, it was predominant in parasites from Sri Lanka, India, Nepal, Pakistan and Madagascar and existed in few parasites from Latin America. In contrast, the haplotype g2 was highly frequent in parasites from Latin America (SMX, Brazil, Honduras and Ecuador). Haplotype g2 was at one mutational step from g5, the latter represent Sal I strain (the only haplotype with codon 958 → acg) which was found in parasites from Latin America and India. Moreover, from g2 emerged a branch of four descendant haplotypes (restricted to American parasites), one present in SMX and NIC-NE (connected to another haplotype present in parasites from SMX, NIC-NE and Brazil), and two exclusive to NIC-NE (Fig. 3). Whereas, haplotype g3 was included in one parasite from Brazil and another from NIC-NW, is at one mutational step from g4, the latter of which showed high frequency in NIC-NE and NIC-NW parasites and was found in one parasite from Brazil (Fig. 3). There were two different haplotypes (g1 and g6) showing polymorphisms T958M and F1076L. Haplotype g6 exclusive in NIC-NE emerged from a different branch of that detected in one parasite from Brazil and other geographic sites outside Latin America (g1).

F _ST values were mostly from low to moderate between sites inside Latin America, excluding NIC-NE and NIC-NW. F _ST values, between SMX, Brazil, Honduras and Ecuador ranged from 0.015 to 0.194. The lowest value was between SMX and Brazil, and the highest between Honduras and Ecuador. In Asia (mostly the Indian subcontinent: Nepal, Iran, Pakistan, India and Sri Lanka), F _ST values were from low to moderate (Table 3). Between NIC-NE and NIC-NW the value was also low (0.097), but higher between these sites and others inside and outside Latin America (0.4784–0.8039). In general, between Latin American and other sites outside this continent, F _ST values were generally high (0.3813–0.9430).